- My News
- All News
- Most Popular
Alphabet's experimental research group X on Monday unveiled insights from an early-stage mental health initiative, dubbed Project Amber, at the Sapien Labs Symposium. The project combined machine learning techniques with electroencephalography (EEG) to develop an objective measurement of depression.
"Amber's small team of neuroscientists, hardware and software engineers, machine learning researchers and med-tech product experts have been developing prototype technologies to help tackle the huge and growing problem of mental health," said Obi Felten, who is head of getting moonshots ready for contact with the real world at X, the moonshot factory. "After three years of exploration, we recently wrapped up our work at X. Now we are making our technology and research findings freely available in the hope that the mental health community can build upon our work," she added.
Felten said X researchers were inspired by work into the brain's reward system. Specific game-like tasks were designed and brain responses were measured using EEG. "It turns out that the brain response following a win in the game," referred to as an event-related potential (ERP), "is subdued in people who are depressed, compared to those who are not," she said. Project Amber focused on making EEG data both easier to collect and interpret, and also on trying to understand how this technology might be applied in the real world.
Researchers at X set out to build an "easy-to-use, low-cost, portable, research-grade EEG system," and came up with a final prototype consisting of a headset that "slips on like a swim cap," and takes around three minutes to set up by anyone with minimal training, according to Felten. The device uses three dry sensors arranged on the scalp at the most important channels for ERP assessments of reward and cognitive function, while the accompanying bioamplifier makes it possible to connect a standard headset with some modifications. Felten noted that the Amber system can be used to collect both resting state EEG and ERPs via software that time-locks a task to the EEG measurement.
The team also explored how new approaches in machine learning could be applied to interpreting EEG data outside electrophysiology laboratories and neurology clinics. Collaborating with DeepMind, the researchers first demonstrated that so-called "representation learning approaches such as autoencoders could be leveraged to effectively de-noise EEG signals without a human EEG expert in the loop," Felten explained. They next offered a proof of concept that it is possible to extract interpretable features that are relevant to mental health.
"We used these features obtained from disentangling autoencoders to predict several clinical labels such as major depressive disorder and generalised anxiety disorder, based on a clinical interview by a mental health expert," Felten said, adding "the methods were capable of recovering usable signal representations from single EEG trials. This means that it may be possible to derive clinically useful information from brain electrophysiology with far fewer data samples than what is traditionally used in research labs."
Over the course of Project Amber, researchers conducted over 250 interviews with potential users of the technology. "More research is needed to determine how a tool such as EEG would be best deployed in clinical and counselling settings, including how it could be combined with other measurement technologies such as digital phenotyping," Felten remarked.
The company said it is open-sourcing its hardware designs, visualiser and stimulus software of the Project Amber prototype EEG system, while pledging the free use of its patents and applications.
To ensure you don't miss other Top Stories like these and news on key medtech industry developments, sign up for our free daily e-newsletter here.
Did you like this article?